Semantic Segmentation on Gleason
57.7mIoUFlatten Swin + relay tokens
Evaluation results
| Method | Links | |
|---|---|---|
| Flatten Swin + relay tokensBackbone=Flatten Swin, Scaling strategy=MS: 256 + 1024 ↓ 42026.01 | 57.7 | |
| SwinV2 + relay tokensBackbone=SwinV2, Scaling strategy=MS: 256 + 1024 ↓ 42026.01 | 55.5 | |
| GLAM + relay tokensBackbone=GLAM, Scaling strategy=MS: 256 + 1024 ↓ 42026.01 | 54.1 | |
| U-NetBackbone=U-Net, Scaling strategy=SW: 5122026.01 | 53.5 | |
| Flatten SwinBackbone=Flatten Swin, Scaling strategy=SW: 2562026.01 | 50.7 | |
| ViT + relay tokensBackbone=ViT, Scaling strategy=MS: 256 + 1024 ↓ 42026.01 | 49.1 | |
| PvTv2Backbone=PvTv2, Scaling strategy=SW: 2562026.01 | 48.6 | |
| SwinV2Backbone=SwinV2, Scaling strategy=SW: 2562026.01 | 48.1 | |
| GLAMBackbone=GLAM, Scaling strategy=SW: 2562026.01 | 47.7 | |
| DeepLabv3Backbone=DeepLabv3, Scaling strategy=SW: 5122026.01 | 46.3 | |
| xTBackbone=xT, Scaling strategy=SW: 10242026.01 | 46 | |
| ViTBackbone=ViT, Scaling strategy=SW: 2562026.01 | 33.2 | |
| Vision MambaBackbone=Vision Mamba, Scaling strategy=SW: 10242026.01 | 22.1 |